Integrations · Twilio

Twilio voice AI agent

Deploy Nexora Voice on top of your existing Twilio number, Programmable Voice flow, or Twilio Flex contact centre. The agent handles inbound and outbound calls through Twilio's Media Streams API, with sub-second latency, full-duplex turn-taking, and barge-in support. SMS, MMS, and WhatsApp Business through Twilio's APIs all connect to the same agent surface — one identity across every channel your numbers carry.

How it works

Three steps to ship Twilio in your pilot

STEP 1

Point your Twilio number

Update your Twilio Programmable Voice flow to forward inbound calls to the Nexora Media Streams endpoint. Outbound calls dial through your existing Twilio sub-account; no number porting required.

STEP 2

Stream audio bidirectionally

Twilio Media Streams pipes the audio to Nexora over a low-latency WebSocket. STT, LLM orchestration, and TTS all happen in the Nexora plane; your Twilio session sees a clean voice on both ends.

STEP 3

Hand off to Flex when needed

When the conversation needs a human, the agent transfers the live call into your Twilio Flex queue with the transcript and signal events as task attributes. The agent sees what was said before they pick up.

What teams use it for

Common Twilio use cases

  • Inbound voice support on your existing Twilio number with Flex queue handoff when escalation is needed.

  • Outbound collections and notification campaigns dialled through Twilio with the same agent persona across channels.

  • Twilio WhatsApp Business chats and Twilio SMS sessions routed to the same agent for channel-consistent customer experience.

Code preview

Twilio — forward Programmable Voice to Nexora Media Streams

YAML
# Twilio voice webhook returns this TwiML
<Response>
  <Connect>
    <Stream url="wss://stream.nexoralabs.co.nz/voice">
      <Parameter name="tenant" value="acme-prod"/>
      <Parameter name="scenario" value="inbound-support"/>
    </Stream>
  </Connect>
</Response>

Stylised example — not a runnable script. Endpoints, payload shapes, and authentication are documented in the integration runbook shipped during the pilot.

Frequently asked questions

Common questions

Do I need to port my number to Nexora?

No. Nexora runs on top of your existing Twilio numbers — Programmable Voice handles the carrier-side audio, Nexora handles the agent-side intelligence over Media Streams. If you ever decide to move providers, your number stays in your Twilio account; Nexora is the agent layer, not the carrier.

What is the latency of Twilio + Nexora end-to-end?

Typical end-to-end latency is 600-900 ms from customer-spoke to agent-replied, including Twilio Media Streams round-trip, STT, LLM, and TTS. We sit in front of the same fast inference path that powers our direct-SIP deployments. Latency targets are part of the pilot success criteria — we measure it under load before go-live.

Can the agent transfer to a Twilio Flex agent live?

Yes. The agent emits a transfer instruction that routes the call into your Flex queue as a TaskRouter task, with the transcript and signal events attached as task attributes. The Flex agent sees what was said and what the customer wants before they pick up — no "start over" experience.

Does the integration cover Twilio WhatsApp and SMS too?

Yes. Twilio WhatsApp Business and Twilio SMS both connect to the same Nexora agent surface. One persona, consistent across voice, WhatsApp, and SMS. Channel handoff between them is supported — start on SMS, switch to a voice callback, and the agent remembers what was already said.

Ready to wire Twilio into your pilot?

Pilot in 14 days. Live in 30.

A 30-minute scoping call. We bring the Twilio integration patterns. You bring the system credentials and the first use case.

Talk to salesSee product capabilities